ClothDojo: Learned Data Generation and Benchmarking for Bimanual Garment Flattening and Folding

Complex manipulation. Reproducible evaluation.

Paper

500 garments.
Diverse shapes and topologies.

The 500-garment core asset pool, used for generation and evaluation; these are not 500 manually demonstrated garments

Re-rendered for learning.
7,848 RGB trajectories.

FlatteningFolding

One trajectory.
Multiple robot platforms.

Start with the code preview.

CPU quickstart

Score a saved cloth state.

Python 3.11 · uv

git clone --depth 1 https://github.com/clothdojo/clothdojo.github.io.git
cd clothdojo.github.io
uv sync --frozen
uv run --frozen python scripts/score_fold.py

Evaluate a policy ↗

Connect a policy server and evaluate against fixed initial states.

Generate & render ↗

Collect trajectories and export RGB training episodes.

Train an RGB policy ↗

Inspect datasets and train with task-matched model bundles.

Requirements & expected output

The code preview includes a CPU scoring example, one garment, and four initial states per task. Full model bundles, datasets and visual assets are forthcoming.

This example runs on CPU, with no simulator or model checkpoint. It checks a single saved state, not a complete rollout.

Expected output: frame_passes_threshold: true

Generation, rendering and training workflows require the relevant model/data/visual resources. See each guide for availability and requirements.